AUG v3 vs ICE Priority Score — diagnostic vs prioritization
ICE (Impact × Confidence × Ease) is Sean Ellis's 3-factor experiment-prioritization framework. AUG v3 is a 7-factor growth-diagnostic framework. They're often confused as alternatives but they answer different questions and compose nicely.
The fundamental difference
AUG v3 answers: “Where is my product's growth friction RIGHT NOW?” — outputs a 0-100 composite + per-factor breakdown + a single biggest weakness to fix this week.
ICE answers: “Of the experiments I'm considering, which should I run FIRST?” — outputs a relative ranking across candidate ideas using Impact × Confidence × Ease.
AUG diagnoses the patient. ICE picks the treatment. Both matter.
Comparison table
| Dimension | ICE | AUG v3 |
|---|---|---|
| Purpose | Prioritize experiments before running them | Diagnose current growth health across 7 factors |
| Factors | Impact (1-10), Confidence (1-10), Ease (1-10) | 7 (Acq, Act, Eng, Ret, Adv, Mon, Perf), each 1-10 |
| Math | Sum or product across 3 factors; relative ranking | Geometric mean × 10 across 7 factors → 0–100 absolute composite |
| Granularity of output | Ranks N candidate ideas against each other | Single composite score + per-factor breakdown + named weakness |
| When to use | You have a backlog of experiment ideas; need to pick the first | You have a live product; want to know which factor is leaking |
| Subjective vs measured | All 3 inputs are subjective estimates | Performance auto-measured (PSI); 6 others self-report against rubric |
| Time to score | 30 sec per candidate idea | ~60 sec for full product audit |
How to use them together (recommended workflow)
- Run AUG audit on your SaaS → /audit/. Get composite score + named weakest factor.
- Read the deep dive for the weakest factor → /method/. Each factor lists the 10 highest-leverage fixes.
- Apply ICE to the candidate fixes from step 2. Score each on Impact × Confidence × Ease.
- Ship the top-ICE fix. Re-run AUG audit in 30 days to measure the lift. Repeat.
AUG tells you what to work on. ICE tells you where to start within that work. Neither replaces the other.
Where the frameworks borrow from each other
AUG v3 (per the methodology page at /method/) explicitly adopts ICE's discipline of separating confidence from magnitude. When rating an AUG factor 1-10, honest scoring requires acknowledging you might be wrong about your retention numbers (low confidence) vs. having concrete cohort data (high confidence). The framework leaves room for both — a 7/10 with confidence 0.9 is more actionable than a 9/10 with confidence 0.3.
Run your audit
Diagnose first; prioritize second. See your AUG score in 60 seconds.